CFC Ratings Motion 2008-10

[Bob Armstrong]

Posted on Old ChessTalk by Peter Stockhausen:

I e-mailed the following message to all Governors regarding the above, passed, Motion:

Dear fellow Governors and dear Bob,

Now that Motion 2008-10 has passed, I think it appropriate to update you on the application.

1, Bob is confident that the Participation and Bonus Point applications will be effective as of July 1, 2008
2, Bob will determine if the calculations will be done quarterly, monthly or even each update.

No doubt, once he has worked out the details, he will brief you. To my mind the Participation and Bonus points are absolutely critical to reverse our membership position by:

a, Allowing lower rated players, particularily Juniors, to improve their ratings, more commensurate with their improvements in strength, more quickly.
b, Remove any and all deflationary tendencies on a continous basis.

So, all you tournament directors, please use your eagle eyes to spot any issues. Like with all new applications, even with absolutely best efforts, there will be some slip ups. After all, even Bob and his team are merely human.

Best Regards
Peter Stockhausen
Treasurer

PS Our thanks are due, once again, to Bob Gillanders, for being really on top of this.

[Bob Armstrong]

Re: CFC Ratings Motion 2008-10 - Text from GL#9

The motion/comments from the GL #9:

RESULTS OF VOTING:
Motion 2008-10: (Moved/Seconded Peter Stockhausen/Lyle Craver)
Bonus Points in the Rating System
A, Participation Bonus (No Restrictions)
Rated 0000 - 1000 2.00 Points per game played
Rated 1001 - 1800 1.50 Points per game played
Rated 1801 - 2000 1.00 Points per game played
Rated 2001 - 2200 0.75 Points per game played
Rated 2201 + 0.50 Points per game played
B, Result Bonus (Performance Rating must exceed highest CFC Rating ever)
0% - 59% No points
60% - 70% 5 Points
71% - 89% 10 Points
90% + 15 Points
Votes YES: (10) Craft, Craver, Dénommée, Haley, Long, Mallon, Ritchie, Smith, Stockhausen, von
Keitz
Votes NO: (1) Barron
Abstentions: (2) Bluvshtein, Wright
Motion Passes
Lyle Craver: This motion is primarily mathematical in nature and made more acute by the current state
of our membership. In general in any Elo type rating system there are players entering and leaving the
rating pool. Most players improve their chess as a result of participating in tournaments thus increasing
their playing strength. Normally a gain in rating from their initial level accompanies this gain in strength.
When a player leaves the rating pool he/she is generally higher rated than when they entered the pool and
thus when they withdraw from active play (which is NOT the same as withdrawing from the CFC though
it often is) they usually take more rating points with them than when they entered. While new players
make ratings a non-zero-sum “game”, the system is rigged that the average rating generally stays more or
less status. This breaks down when the rating pool shrinks and the average rating declines. This rating
deflation is what this motion is designed to prevent – and any time a rating boon of this sort is proposed
it’s going to be controversial because the issue is not to be too much nor too little but just right. That’s a
judgement call rather than something that can be scientifically determined and this is a reasonable antideflationary
measure.
Michael Barron: I believe before further tinkering with our rating system we need to assess the results of
2006 “rating boon” first. I see more harm than benefit from this boon: tournament participation and CFC
membership continued to decline, and I know at least 2 issues caused by the “boon”:
1) It almost ruined GTCL Under 2000 League 2006-2007. After the ”boon” some active players got more
than 200 extra rating points and overcame 2000 mark. Losers immediately started to whine that such
players couldn’t play in Under 2000 League anymore. Thankfully, GTCL rules clearly state that
eligibility to participate defined by the rating at the start of the season (in October). But if our rules were
not clear enough, half of the teams couldn’t finish this League.
2) One of Toronto players was deprived of his GTCL Grand Prix prize, which he certainly deserves. The
“boon” artificially put him in the next rating class. During 2007 year he played well enough, but anyways
lost some of unearned rating points and returned to his usual rating class. The problem is that Grand Prix
prizes are limited by rating class, which defined by the rating at the start of the series (in March). Because
his rating in March was artificially high, this player was not eligible for the prize in his rating class.
So, I would avoid any changes to the rating system, at least until comprehensive analysis of the recent
changes and trends is done.
The integrity and continuity of the rating system are much more important than any perceived benefits,
especially when we use CFC ratings to determine eligibility for various events and prizes.
Pierre Dénommée: I vote yes for maketing reason, this may attract and retain more players.
Chris Mallon: I vote YES to 2008-10, and would like to add the following comment:
"I suspect that this will end up just like the last change to bonus points we made about a year ago when
the peak rating provision was removed; our current software can no longer be modified since we no
longer have the source code, so this will be another empty change that makes no real difference until we
get new rating software."
Stephen Wright: I wrote this in the previous GL:
"Is there any mathematical/scientific basis for the motion? For that matter, there was a ratings boon some
18 months ago - have there been any studies to determine if this had the desired effect, or whether further
modification (possibly including this motion) is required? Is the rating system being overseen on an
ongoing basis?"
There has been no response - I therefore do not have enough objective information to be able to vote on
this motion. Sorry, but having people say "it's a good idea" without any analytically support is
insufficient justification to modify a ratings system, in my view.

[Bob Armstrong]

Re: CFC Ratings Motion 2008-10 - Text from GL#9

The governors are doing some good things at the end of their mandate. This is a good housekeeping measure, and directly relates to member servicing. Now if we can only get them to pass the critical restructuring motions at the CFC AGM, and get themselves out of the red ink!

[Gary Ruben]

Getting Honest

The governors are doing some good things at the end of their mandate. This is a good housekeeping measure, and directly relates to member servicing. Now if we can only get them to pass the critical restructuring motions at the CFC AGM, and get themselves out of the red ink!

Hi Bob,

Those restructuring motions aren't really a staw vote, are they? If the Governors approve all, or some of them, isn't it expected they will be enacted?

I don't know how else to read a post which call the motions "critical restructuring motions".

[Bob Armstrong]

Re: The Value of Straw Votes

You are right Gary. my enthusiasm is a bit effusive. The 7 motions are indeed straw vote motions. However, although they are not binding, they are critical all the same. If all 7 get passed strongly, this is a clear signal to the new Governors and new Executive, of the restructuring plan they should then immediately implement.

They don't have to, I suppose, but the opinion of the experienced outgoing governors will carry good weight I think ( I hope ).

Bob

[Jason Lohner]

Re: CFC Ratings Motion 2008-10 - Text from GL#9

The motion/comments from the GL #9:

Stephen Wright: I wrote this in the previous GL:
"Is there any mathematical/scientific basis for the motion? For that matter, there was a ratings boon some
18 months ago - have there been any studies to determine if this had the desired effect, or whether further
modification (possibly including this motion) is required? Is the rating system being overseen on an
ongoing basis?"

The voice of common sense. Why should there be more 'tinkering' with the ratings formula without a specific reason for it. If there is an actual mathematical/scientific reason for increasing ratings then do it.

[Jean Hébert]

Re: CFC Ratings Motion 2008-10 - Text from GL#9

This motion adresses the symptom of the problem but not the root of it. Too many high rated players are quitting and no "mathematical" motion will lessen this problem. When a pool of players loses the top ones, it also loses much of its capacity to develop replacements. To move up the rating ladder you need opponents with higher ratings. It would be more useful to find out means to keep those higher rated players into the game!

[Jonathan Berry]

Re: CFC Ratings Motion 2008-10 - Text from GL#9

In the ensuing discussion I'll use the old 16/4%, 8/2% rating system. If it turns out that equivalent examples can't be made with the actual rating system, I apologize in advance.

The motion/comments from the GL #9:

RESULTS OF VOTING:
Motion 2008-10: (Moved/Seconded Peter Stockhausen/Lyle Craver)
Bonus Points in the Rating System
A, Participation Bonus (No Restrictions)
Rated 0000 - 1000 2.00 Points per game played
Rated 1001 - 1800 1.50 Points per game played
Rated 1801 - 2000 1.00 Points per game played
Rated 2001 - 2200 0.75 Points per game played
Rated 2201 + 0.50 Points per game played

This adds rating points to the system faster than the 1 point per game anti-deflationary measure (known as Participation Points) of decades ago. Demographics alone (the population in general is older than in 1980) would tell us that deflation is less than it was then. On the other hand, reportedly there are more youth players as a proportion of the tournament population, and some are coached rather than self-taught, as in days of yore.

On balance, I'd say that if you're not willing to do the research (which was done for the original participation points, even if its "point" was more the point of a pencil than the program of a computer), this is not an unreasonable measure. In an arm-waving, easychair executive kind of a way. I come not to bury Caesar, but to chuckle a bit.

Research might involve this: examine the ratings of players whose chess strengths you'd expect to be staying about the same. Following Arpad Elo, you might select players aged 30-40. The birthdates are in the CFC database, though not in the TDLIST available to the public, for good reason. If a significant random sample of such players shows that they are losing on average say 15 points per year, the purpose of an anti-deflationary measure would be to inject back 15 points for every player in the system. BTW, AFAIK, Robert Hamilton did not do this sort of research for the recent Rating Boon, although it was suggested to him.

Note that the Plan A points are highly redistributional. If a player rated 1850 becomes 25 points overrated relative to the pool, then for every game against an opponent within range, his participation point is nullified by the 4% of the opponent's rating.

Note also that by 1986 (mag 80, page 62) the Participation Point measure excluded players (all those rated over 2400) who might qualify for the Olympiad teams. It might be a good thing, but it does change the slope of the pitch. Also, the points gained by 2500 players are less redistributional than those of the 1850 player because of the 2% and also because they are more likely to meet opponents who are out of range.

B, Result Bonus (Performance Rating must exceed highest CFC Rating ever)
0% - 59% No points
60% - 70% 5 Points
71% - 89% 10 Points
90% + 15 Points

I confess that I skipped over this section, Plan B. I didn't realize it existed until it was mentioned by Ben at the old ChessTalk.
http://members5.boardhost.com/ChessT...215903465.html
(until it scrolls off, forever!)

LOL. Viewing the approval of Plan B in the kindest possible way, it is a good sign. In recent months, the CFC had (at least) two horrible decisions. The CFC Governors, in their wisdom, showed that they were more interested in the more important of these two horrible acts, the one dealing with the redistribution of CFC's monetary assets rather than of its rating point assets. Thus the Governors bucked Parkinson's Law, which requires that attention be applied in inverse proportion to the importance of a matter. Congratulations, CFC Governors!

Rating examples: Let's say that a 1600 player scores 3-2 against a quintet of 1900-rated opponents. He gains about 16 + (5x12) = 76 points (ignoring the participation points from part A), plus 5 bonus points, so his new rating is 1681. That is, about 300 points below his performance in the tournament. Under the old (1980) dispensation, he would have gained 76-26=50 bonus points, and his new rating of 1726 somewhat closer to his performance. In short, Plan B is stingy to the rapidly-improving player.

Case 2. A player rated 2500 scores 4.5 against a quintet of 2181-rated opponents. He gains 32 - 31 + 15 = 16 points, so his new rating is 2516, some 15 points higher than his performance in the tournament. And, just for fun, this new rating qualifies him for the Olympiad team ahead of the guy who tied for first with him, the guy whose pre-event rating was 2505 and played the same opposition.

Case 3. A match.

Note that the Plan B points are not so redistributional as the Plan A points.

By or before 1986 (mag 80, page 62), players rated over 2300 were excluded from bonus point schemes, so they didn't affect Olympiad team selection.

In brief, Plan B is deeply flawed. It probably won't generate a great number of points, but when it does, it will award them inequitably. It also carries the possibility of skewing national team selection.

In this discussion:
http://members5.boardhost.com/ChessT...215883448.html
on the old ChessTalk (until it scrolls off forever!) Peter Stockhausen stated (or at least strongly implied) that the measures of this motion, including Plan B, are the same as those used earlier in the CFC rating system. They aren't the same as the Bonus Point system used for most of the last two decades of the 20th century (see magazine #42, pp 45 ff or mag #80, p.62). I'd like to see the exact wording of a plan B equivalent from later in the life of the CFC Rating System, but don't know where to look it up. I do have all the CFC magazines ... any pointers? I suppose that if the exact wording of plan B was used before, it serves to shift the blame to some earlier Assembly of Governors. It does not make the measure any sounder.

The 2007-08 Governors were distracted by promises of forthcoming reports on that other matter. But the Rating Auditor might have taken his duty to "supervise the operation of the rating system" on this, and on a few other occasions in recent years, as an invitation to try to keep it working sensibly.

Votes YES: (10) Craft, Craver, Dénommée, Haley, Long, Mallon, Ritchie, Smith, Stockhausen, von
Keitz
Votes NO: (1) Barron
Abstentions: (2) Bluvshtein, Wright
Motion Passes

[Bob Gillanders]

Re: CFC Ratings Motion 2008-10

I have crunched some stats on ratings from the CFC membership database. This data is for current(at that time) adult members only. I picked this group as being the most stable subgroup of members. FYI - I have been downloading member list on a regular basis over the last year. My first data point is from Nov 28, 2007 because data prior to that date is unreliable. Members will ratings of zero for whatever reason are excluded.

Here's the data

date,...... # of members,... average rating

Nov 28, 2007.......837.......1,722
March 12, 2008....842.......1,718
May 1, 2008........912.......1,717
May 22, 2008.......914.......1,712
July 1, 2008.........914.......1,712

The data would tend to support the theory of rating deflation, but I make no claim as to it's statistical significance. Just the facts, Jack.

A comparison to 10, 20, and 30 years ago would be interesting !!

With this baseline data, we can track the effect of motion 2008-10.

[Jonathan Berry]

Re: CFC Ratings Motion 2008-10

Thanks, Bob. Although it's nice to see some figures, any figures--players over 40 are supposed to get weaker. If more of the active tournament participations are by players aged 41-999 than 18-29, then it is not clear that there is any rating deflation. I'd also exclude provisional ratings. I wonder what the figures are if you select just the players aged from 30 to 40.

But as a PR exercise for the CFC, I think that Plan A, or something like it, is mostly OK. It's Plan B that raises doubts.